Skip to content
All library documents

Few-Shot Learning for Adaptive Trend-Following Forecasts

Article arXiv papers · Author: Kieran Wood et al.
Read the original

Summary

The document proposes X-Trend, a trend-following time-series forecaster designed to adapt quickly when markets enter new regimes. It uses few-shot learning and cross-attention over a context set of financial time-series regimes, transferring patterns from similar regimes to forecast positions in a target regime. The attention mechanism is also presented as a way to interpret which context patterns relate to forecasts.

The reported evaluation covers a turbulent period from 2018 to 2023. The authors state that X-Trend improved Sharpe ratio over a neural forecaster and a conventional time-series momentum strategy, recovered faster from the COVID-19 drawdown, and produced positions for unseen assets with reported Sharpe gains over a neural forecaster. These are results reported by the document; it gives no details here on data construction, transaction costs, statistical uncertainty, or out-of-sample safeguards, which limits independent assessment.

Key ideas

  • X-Trend uses few-shot learning to adapt trend forecasts to new market regimes.
  • Cross-attention transfers patterns from related regimes in a context set to a target regime.
  • The authors report Sharpe improvements over neural and conventional momentum benchmarks during 2018–2023.
  • The approach is reported to recover faster from the COVID-19 drawdown and handle unseen assets.
  • Cross-attention provides a way to inspect relationships between context patterns and forecasts.

Tags

Trend followingMachine learningMomentumBacktesting
Full text
# Few-Shot Learning Patterns in Financial Time-Series for Trend-Following Strategies


# Few-Shot Learning Patterns in Financial Time-Series for Trend-Following Strategies









Forecasting models for systematic trading strategies do not adapt quickly when financial market conditions rapidly change, as was seen in the advent of the COVID-19 pandemic in 2020, causing many forecasting models to take loss-making positions. To deal with such situations, we propose a novel time-series trend-following forecaster that can quickly adapt to new market conditions, referred to as regimes. We leverage recent developments from the deep learning community and use few-shot learning. We propose the Cross Attentive Time-Series Trend Network -- X-Trend -- which takes positions attending over a context set of financial time-series regimes. X-Trend transfers trends from similar patterns in the context set to make forecasts, then subsequently takes positions for a new distinct target regime. By quickly adapting to new financial regimes, X-Trend increases Sharpe ratio by 18.9% over a neural forecaster and 10-fold over a conventional Time-series Momentum strategy during the turbulent market period from 2018 to 2023. Our strategy recovers twice as quickly from the COVID-19 drawdown compared to the neural-forecaster. X-Trend can also take zero-shot positions on novel unseen financial assets obtaining a 5-fold Sharpe ratio increase versus a neural time-series trend forecaster over the same period. Furthermore, the cross-attention mechanism allows us to interpret the relationship between forecasts and patterns in the context set.

Shown in full with attribution under the source's licence. Licence: abstract CC0

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.